
In April 2023, University of Louisville eating disorder researcher Dr. Cheri Levinson told STATnews, “I don’t think that there should be any form of treatment for obesity.” She acknowledged in the same interview how extreme she considered her position. Three years later, she asks whether weight stigma has clouded scientific judgment about GLP-1 drugs. Before accepting her judgment about the FDA’s scientific judgement, readers should examine how she handles evidence that challenges her own position.
Start with bariatric surgery. In 2021, a meta-analysis involving 174,772 participants found that people who underwent surgery lived longer on average than those who received nonsurgical care. By January 2024, when Dr. Levinson wrote about Ozempic for her laboratory’s website, that finding had been public for nearly three years. Yet she grouped surgery with Weight Watchers and intermittent fasting in a discussion of weight-loss approaches that fall out of fashion or are debunked. Inexplicably, she did give space to adverse experiences reported in a mothers’ social-media group. Which leads us to question, where was the comparable attention to the survival evidence?
Dr. Levinson has made clear that she supports treating diseases such as diabetes and hypertension but opposes weight loss itself as a treatment objective. She is entitled to that position. However, what she cannot do is use concerns about eating disorders to advance broader warnings about GLP-1 medications without acknowledging the substantial clinical evidence of their benefits. These medications have demonstrated reductions in serious cardiovascular events and in certain patient populations, mortality.
Those findings do not eliminate legitimate concerns about eating disorder risks. They do, however, expose the danger of presenting a selective account of the science. When a researcher publicly questions the safety of a medication, intellectual honesty requires acknowledging evidence that challenges the narrative, not merely the evidence that supports it.
That history gives readers a concrete reason to scrutinize her recent LinkedIn post. Dr. Levinson places 31 U.S. deaths linked to Baycol, a cholesterol drug withdrawn in 2001, alongside 4,572 reports listing death among people taking GLP-1 medications. She acknowledges that the reports cannot prove causation, then asks where the comparable regulatory scrutiny is. The credibility of that challenge turns on two questions her comparison leaves unanswered. What did investigators know about the Baycol deaths, and what has the FDA actually investigated about GLP-1 drugs?
What the Two Death Counts Actually Measure
Bayer withdrew Baycol after reports of rhabdomyolysis, a severe breakdown of muscle tissue that can cause kidney failure and death. Investigators could name the injury, count prescriptions and compare reports involving other statins. The reporting rate for fatal rhabdomyolysis with Baycol was approximately 16 to 80 times the rate for other statins. Higher doses and concurrent use of gemfibrozil gave investigators further clues about which patients faced the greatest danger. Thirty-one deaths did not constitute a universal threshold for pulling a medicine from the market. The pattern behind those deaths made the case.
Dr. Levinson’s utilizing 4,572 deaths pools four medications over twelve calendar years. However, she did not ask what conditions caused those deaths. Or, how many people took each medication, for how long, and how many deaths would ordinarily occur among comparable patients with diabetes, kidney disease or cardiovascular disease? Without that information, even a perfectly accurate count of 4,572 cannot tell us whether the use of GLP-1 medication increased mortality. In fact, the FDA expressly warns against using its reporting system to calculate incidence or infer causation. An acknowledgment of that limitation carries little weight if the argument then asks readers to make the very inference the limitation rules out.
A published analysis illustrates the problem. Through February 17, 2026, researchers counted 64,333 adverse-event reports involving semaglutide. Of those, 33,499 were classified as serious, including 949 that listed death. For tirzepatide, 19,971 of 127,005 reports were serious, including 633 that listed death. Those figures cover all types of reported events. They neither reproduce Dr. Levinson’s four-drug tally nor establish that either medicine caused a particular death. They certainly cannot tell a parent how often an eating disorder develops or worsens during treatment.
There is a straightforward way to put the 4,572 figure on a more substantive footing. Publish the search dates and drug names, explain how duplicate and follow-up reports were handled, and distinguish cases identifying a drug as a suspect from those merely listing it among a patient’s medications. That would allow another researcher to reproduce the count. Even with that additional research, establishing excess mortality would still require a comparison with patients of similar age, illness and baseline risk.
The FDA Investigations Already on the Record
On November 9, 2026, the FDA required a new warning about a rare optic nerve injury that can cause permanent vision loss. Its analysis of 126 randomized trials found 6.4 cases per 100,000 person-years among patients receiving GLP-1 drugs, compared with 3.8 among patients receiving placebo or another glucose-lowering medication. the FDA acting in its role as the regulator publicly identifying an injury, estimating its frequency in two groups, acknowledging uncertainty and directing manufacturers to change their labels. Perhaps, a challenge to the adequacy of FDA scrutiny should have confronted that decision.
The FDA also examined reports of suicidal thoughts and behavior. In January 2026 it announced that its review found no increased risk and asked manufacturers to remove that warning from the labels for Saxenda, Wegovy and Zepbound. Across these two investigations the FDA strengthened one warning and called for removal of another.
Anyone claiming that weight stigma explains the agency’s conduct owes readers evidence that it distorted the FDA’s investigations into GLP-1 medications. The mere fact that the drugs remain available supplies none.
The Diagnosis that Disappears inside the Phrase “Eating Disorders.”
We should also consider Dr. Levinson’s claim that adolescents in larger bodies are more likely to have “eating disorders.” Indeed, a Canadian community study found that boys in the obesity category had nearly eight times the odds of meeting questionnaire criteria for a bulimic disorder as boys in the normal-weight category. For girls, the odds were about three times as high. That is an important finding and an even greater distinction.
Of great importance, Dr. Levinson changed the term, “bulimic disorders” to “eating disorders.” This completely changes the claim and analysis. It invites a reader to erroneously apply the same data and statistics to anorexia nervosa, a conclusion the study did not establish. This intentionally omits the eating disorder with the highest mortality rate. Dr. Levinson’s claim is disingenuous at best. It also calls into question Dr. Levinson’s statement, “I don’t think that there should be any form of treatment for obesity.” If that statement still applies, can we presume that Dr. Levinson does not support any form of treatment for the obese boys and girls in the Canadian community? And simply allow them to suffer the pain and indignities of bulimia?
To further address Dr. Levinson’s claims, a British study following children over time found a different pattern among those who later developed anorexia nervosa. Their childhood BMI was lower on average than that of their peers. Meanwhile, clinical research has also documented serious medical complications in adolescents with atypical anorexia, including those who began at higher weights. An updated mortality review found the highest standardized mortality ratio among eating disorders is still to be found in anorexia nervosa. Dr. Levinson did not address that most important statistic. For that matter, very few fat activists do. That reality is too inconvenient. It is too … real.
This is where selective advocacy does damage to the public understanding of eating disorder research. A study carefully distinguishes diagnoses and reports a finding about one of them. In the retelling, the qualification vanishes while the authority of the study survives. Families hear that researchers have settled a question those researchers took considerable care to define more narrowly. The damage is visible in the claim itself. Whether any particular patient changed treatment after reading a post is a separate question.
What the Field Should Demand
Transparency. Accuracy. Integrity. Agenda-free research. Not personal agendas nor tribal viewpoints.
There is certainly a research agenda here worth defending. Follow adolescents with restrictive symptoms who start an appetite-suppressing medication. Document their diagnoses, growth histories and reasons for treatment. Compare outcomes with those of clinically similar patients receiving other care. Record relapse, medical instability and hospitalization alongside any benefits. Tell families where trial exclusions leave a gap. Those are questions upon which supporters and critics of these drugs should be able to collaboratively ask.
Dr. Levinson asks whether weight stigma is clouding scientific judgment. Yet her own argument raises precisely that concern. She places 4,572 deaths alongside 31 without establishing that the figures represent comparable causes of death or accounting for the factors necessary to make a scientifically defensible comparison. She questions the scrutiny directed at GLP-1 drugs while failing to address the FDA’s 20n26 findings about GLP-1 medications. She also treats “eating disorders” as a single, interchangeable category, disregarding critical differences in diagnosis, severity, and mortality.
These are not minor omissions. They go directly to the lack of credibility of her argument and raise a fundamental question: Is Dr. Levinson objectively evaluating the scientific evidence, or selectively presenting it to advance an ideological position she has already embraced?
Dr. Levinson is not being asked to establish that GLP-1 medications are harmless. She is being asked to substantiate the claims she has chosen to make publicly. That requires more than placing two dramatically different mortality figures side by side without establishing their comparability or explaining what they actually measure. It requires distinguishing documented adverse events from deaths causally attributable to a medication. And it requires recognizing that eating disorders are distinct psychiatric illnesses with different clinical characteristics, risks, and mortality rates. These are elementary obligations of scientific accuracy, particularly for a researcher whose professional authority lends credibility to her public statements. When those distinctions are ignored, advocacy risks displacing the very scientific judgment it purports to defend.
American Psychological Association’s guidance on public use of social media asks psychologists to consider the scientific basis for the information they share. An eating disorder researcher earns public authority by making distinctions that a frightened parent cannot reasonably be expected to make alone. When a researcher blurs those distinctions in service of a position on obesity treatment, careful studies are made to say things their authors never established. Families are left to sort out the consequences.
If Cheri Levinson wants to challenge the scientific judgment of others, she should begin by applying the same standards to her own public claims.
















